LOO: Leave-one-out analysis for quantitative evaluation of the...
In snaketron/test_repo: Differential gene usage in immune repertoires

Description Usage Arguments Details Value Author(s) See Also Examples

View source: R/Loo.R

IgGeneUsage detects differential gene usage in immune repertoires that belong to two biological conditions with its function DGU. To assert quantitatively the robustness of the estimated probability of DGU (pi), IgGeneUsage has a built-in procedure for a fully Bayesian leave-one-out (LOO) analysis. During each step of LOO, we discard the data of one of the repertoires, and use the remaining data to analyze for DGU with IgGeneUsage. In each step we recorded pi for all genes. Therefore, by evaluating the variability of pi for a given gene, we can we assert quantitatively its robustness.

Notice, however, that for datasets that include many repertoires (e.g. 100) LOO can be computationally costly.

1
2
3

LOO(usage.data, mcmc.warmup, mcmc.steps,
    mcmc.chains, mcmc.cores, hdi.level,
    adapt.delta, max.treedepth)

`usage.data`	Data.frame with 4 columns: 'sample_id' = character identifier of each repertoire, 'condition' = character key representing each of the two biological conditions, 'gene_name' = character name of each gene to be tested for differential usage, 'gene_usage_count' = number of rearrangements belonging to a specific sample_id x condition x gene_name. Alternatively, usage.data can be a SummarizedExperiment object. See examplary data 'data(Ig_SE)' for more information.
`mcmc.chains, mcmc.warmup, mcmc.steps, mcmc.cores`	Number of MCMC chains (default = 4), number of cores to use (default = 1), length of MCMC chains (default = 1,500), length of adaptive part of MCMC chains (default = 500).
`hdi.level`	Highest density interval (HDI) (default = 0.95).
`adapt.delta`	MCMC setting (default = 0.95).
`max.treedepth`	MCMC setting (default = 12).

IgGeneUsage invokes the function DGU in each LOO step. For more details see help for DGU or vignette 'User Manual: IgGeneUsage'.

loo.summary

differential gene usage statistics for each gene of a given LOO step. 1) es = effect size on differential gene usage (mean, median standard error (se), standard deviation (sd), L (low boundary of HDI), H (high boundary of HDI); 2) contrast = direction of the effect; 3) pmax = probability of differential gene usage; 4) loo.id (LOO step ID); 5 Neff (effective sample size), Rhat (potential scale reduction factor)

Simo Kitanovski <simo.kitanovski@uni-due.de>

DGU, Ig, IGHV_Epitopes, IGHV_HCV, Ig_SE

## Not run:

# input data
data(Ig)
head(Ig)

# Alternative:
# use SummarizedExperiment input data
data(Ig_SE)


# run leave-one-out (LOO)
L <- LOO(usage.data = Ig,
         mcmc.warmup = 250,
         mcmc.steps = 1000,
         mcmc.chains = 2,
         mcmc.cores = 1,
         hdi.level = 0.95,
         adapt.delta = 0.95,
         max.treedepth = 13)

## End(Not run)